Foglight Application Performance Monitoring
Last week a couple of intrepid folks from our Application Performance Management and Cloud Management groups attended the DevOpsDays event in Austin, TX. Barton George, Director of Cloud Developer programs at Dell, conducted a series of video interviews with keynote speakers and other interesting personalities he encountered there.
This was the first DevOps Days show I had attended. Overall, I have say I was pleasantly surprised. The venue was different from a normal conference or trade-show, held in an old theater, but it fit the theme and worked well. It ran more like a user-group than a trade-show, and was very fast paced.
They had Ignite sessions in the afternoons where the presenter was allowed 20 slides and they auto-advanced every 15 seconds, so no droning on, they got to the point and moved fast. Done in 5 minutes and the next guy would come up. It was the first time I had seen that format, and I have to say I liked it. Great way to get an overview of a topic, and you could always contact the presenter for more details if you wanted.
People were there to share ideas and understand how their colleagues were trying to solve similar problems. Capturing data and analytics seemed pretty important to the group. They were data hungry. One presenter summed it up with “more data is nearly always a good thing.” Even if they didn’t look at it for months, they liked the idea of having it to go back to, which makes sense.
Automation topics were well-attended. A lot of discussion on various ways to use Chef/Puppet in new and interesting ways. The thinking seemed to be if you get the automation in place it makes everything else so much easier, from fault recovery to auto-scaling. Makes a lot of sense. One of the presentations covered “Resilience over strength” meaning it is better to have flexibility and automation in your deployment and ops than trying to get to “strength” where an instance never has a problem. That way if you inadvertently introduce a problem, your automation and autoscaling protect you from it. Along the same lines, there were discussions about spending time optimizing the start-up time of your image. The quicker it came up, the more responsive you were for elastic events. If you have an image that takes a long time to boot up and start taking requests, you have to be much more conservative in your scaling rules to make sure you do not get behind the load curve waiting on startup.
I thoroughly enjoyed the two days. I am signed up and looking forward to DevOps Days in Santa Clara June 27-28!